Instructions to use LiquidAI/d1-2B-MLX-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LiquidAI/d1-2B-MLX-4bit with MLX:

# Make sure mlx-vlm is installed
# pip install --upgrade mlx-vlm

from mlx_vlm import load, generate
from mlx_vlm.prompt_utils import apply_chat_template
from mlx_vlm.utils import load_config

# Load the model
model, processor = load("LiquidAI/d1-2B-MLX-4bit")
config = load_config("LiquidAI/d1-2B-MLX-4bit")

# Prepare input
image = ["http://images.cocodataset.org/val2017/000000039769.jpg"]
prompt = "Describe this image."

# Apply chat template
formatted_prompt = apply_chat_template(
    processor, config, prompt, num_images=1
)

# Generate output
output = generate(model, processor, formatted_prompt, image)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use LiquidAI/d1-2B-MLX-4bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "LiquidAI/d1-2B-MLX-4bit"

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "LiquidAI/d1-2B-MLX-4bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use LiquidAI/d1-2B-MLX-4bit with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "LiquidAI/d1-2B-MLX-4bit"

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default LiquidAI/d1-2B-MLX-4bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use LiquidAI/d1-2B-MLX-4bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "LiquidAI/d1-2B-MLX-4bit"

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "LiquidAI/d1-2B-MLX-4bit" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

MLX-VLM support?

by dalisoft - opened 10 days ago

Discussion

dalisoft

10 days ago

Hi @LiquidAI team

thank you for small, fast and smart decision model.
As i see https://github.com/Blaizzy/mlx-vlm/pull/2611 is merged but this model card (MLX-4bit) is not updated. Also, LM Studio says not supported for this model (maybe due of vision/mlx-vlm).

Can you fix or help me please?
Thank you

ykhrustalev

Liquid AI org 8 days ago

•

edited 8 days ago

Hugging Face CLI

Hi, thanks for the report.

d1 support shipped in mlx-vlm 0.7.11 (https://github.com/Blaizzy/mlx-vlm/pull/2611). With pip install -U "mlx-vlm>=0.7.11" the example on this card runs as written and prints the answers shown there. We updated this card and the other d1-2B and d1-Omni-2B MLX cards to say so.

LM Studio has two separate problems with this model:

LM Studio runs MLX models through https://github.com/lmstudio-ai/mlx-engine, which currently pins mlx-vlm 0.7.3. That version predates LFM3, so it cannot load any lfm3_vl model, d1-2B included.
Even once LM Studio updates its mlx-vlm, d1-2B will not work as a chat model. d1-2B is a decision model: it answers named, typed questions with calibrated probabilities in one forward pass instead of generating text, so mlx-vlm refuses chat requests for it on purpose.

On a Mac you can run it today from Python with model.system_one(...), as in the card, or from the command line:

pip install -U "mlx-vlm>=0.7.11"
python -m mlx_vlm decide --model LiquidAI/d1-2B-MLX-4bit \
  --state "I was charged twice this month. Please refund one of the charges." \
  --questions '{"refund": {"type": "noul", "instructions": "Is the customer asking for a refund?"}}'

dalisoft

8 days ago

Thank you for detailed response. Yes, directly it works

dalisoft changed discussion status to closed 8 days ago

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment